Blind Denoising of SAR Ship Images Based on Improved Threshold Wavelet Transform*
Jian Yang, Min He, Tongzhou Han, Yanjun Li · 2025
This paper addresses the issue of severe speckle and additive noise interference in Synthetic Aperture Radar (SAR) imaging of maritime ship targets. An improved threshold wavelet-based SAR image denoising method is proposed. For SAR images, the noise type is not assumed to be known a priori, and wavelet denoising is performed in the azimuth direction, enabling more effective optimization based on the specific noise distribution pattern of the current azimuth. After wavelet decomposition, the improved threshold function proposed in this paper is used for denoising processing. The thresholding function introduces a negative exponential decay factor, which allows for adaptive adjustment of the threshold size according to the noise estimation in different azimuth rows. This approach prevents excessive denoising when the noise intensity is high and incomplete denoising when the noise intensity is low. Experimental results show that compared with the traditional threshold wavelet denoising algorithm and SAR-BM3D algorithm, the improved algorithm proposed in this paper has higher denoising performance and image reconstruction quality in the process of SAR ship image denoising.